Gene Fusion, Fission, Lateral Transfer, and Loss: Not-So-Rare Events in the Evolution of Eukaryotic ATP Citrate Lyase
Total Page:16
File Type:pdf, Size:1020Kb
Molecular Phylogenetics and Evolution 91 (2015) 12–16 Contents lists available at ScienceDirect Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev Short Communication Gene fusion, fission, lateral transfer, and loss: Not-so-rare events in the evolution of eukaryotic ATP citrate lyase q ⇑ Ryan M.R. Gawryluk 1,2, Laura Eme 1, Andrew J. Roger Centre for Comparative Genomics and Evolutionary Bioinformatics, Department of Biochemistry and Molecular Biology, Dalhousie University, 5850 College Street, Halifax, Nova Scotia B3H 4R2, Canada article info abstract Article history: ATP citrate lyase (ACL) is an enzyme critical to the generation of cytosolic acetyl-CoA in eukaryotes. In Received 17 December 2014 most studied organisms, ACL activity is conferred in combination by two proteins, ACLA and ACLB Revised 7 May 2015 (dsACL); however, animals encode a single-subunit ACL (ssACL) – the result of a gene fusion event. Accepted 11 May 2015 Through phylogenetic analyses, we investigated the evolution of ACL in a broad range of eukaryotes, Available online 27 May 2015 including numerous microbes (protists). We show that the fused form is not restricted to animals, and is instead widely distributed among eukaryotes. Furthermore, ssACL and dsACL are patchily distributed Keywords: and appear to be mutually exclusive; both types arose early in eukaryotic evolution. Finally, we present ATP citrate lyase several compelling hypotheses of lateral gene transfer and gene loss, along with the secondary gene Gene fusion Gene fission fission of ssACL in Ascomycota. Collectively, our in-depth analyses suggest that a complex suite of Lateral gene transfer evolutionary events, usually considered rare, has shaped the evolution of ACL in eukaryotes. Eukaryotic evolution Ó 2015 Elsevier Inc. All rights reserved. Protist 1. Introduction enzyme that catalyzes the ATP-dependent cleavage of citrate into oxaloacetate and acetyl-CoA. Typically, ACL’s substrate is Acetyl coenzyme A (acetyl-CoA) is a high-energy metabolite that mitochondrion-derived citrate; this enzyme therefore plays a role is a product of carbohydrate, amino acid, and lipid catabolism, and in the ‘citrate shuttle’ that effects the net transfer of acetyl-CoA the precursor of numerous anabolic pathways (Oliver et al., 2009). equivalents to the cytosol for fatty acid biosynthesis. The Given its role at the ‘hub’ of cellular metabolism, a thorough under- acetyl-CoA generated by ACL is a key substrate of myriad other standing of acetyl-CoA biosynthesis is critical; this is especially true downstream anabolic processes in eukaryotes, including the of eukaryotes, as acetyl-CoA is membrane-impermeable, and dis- biosynthesis of sterols, waxes, isoprenoids, and flavonoids (Oliver tinct biosynthetic mechanisms are therefore required in the various et al., 2009), and nuclear histone acetylation (Wellen et al., 2009). subcellular compartments, including mitochondria, chloroplasts, ACL is also encoded in the genomes of prokaryotes; however, it peroxisomes, and the cytosol (Oliver et al., 2009). is sparsely distributed, and is found only in a few species belonging In animals (Elshourbagy et al., 1990, 1992), land plants (Fatland to e-proteobacteria, Aquificae, Chlorobi, and Euryarchaeota, many et al., 2002), a glaucophyte alga (Ma et al., 2001), and filamentous of which are thermophiles living near deep-sea vents (Campbell fungi (Hynes and Murray, 2010; Son et al., 2011), the major cytoso- and Cary, 2004). Although the chemical reaction catalyzed by lic source of acetyl-CoA is ATP-citrate lyase (ACL; EC 2.3.3.8), an ACL is the same as in eukaryotes, the physiological context is dif- ferent: in prokaryotes, ACL is a component of the reverse TCA cycle, a reductive, carbon-fixing pathway that serves as an alternative to Abbreviations: ACL, ATP citrate lyase; ssACL, single-subunit ACL; dsACL, the Calvin–Benson–Bassham reductive pentose phosphate cycle dual-subunit ACL; ACLA, dsACL protein homologous to the N-terminal portion of (Buchanan and Arnon, 1990). Thus, there has been a functional ssACL; ACLB, dsACL protein homologous to the C-terminal portion of ssACL; CS, citrate synthase; SCS, succinyl-CoA synthetase; ML, Maximum Likelihood; LGT, modification in eukaryotes, with ACL shifting from a role in permit- lateral gene transfer; EF 1-alpha, elongation factor 1-alpha; EF-L, elongation ting autotrophic growth, to supplying a key intermediate in various factor-like. eukaryotic anabolic processes (Fatland et al., 2002). q This paper was edited by the Associate Editor Austin Hughes. In bacteria (Hügler et al., 2007; Kanao et al., 2001), a glauco- ⇑ Corresponding author. phyte alga (Ma et al., 2001), green algae/land plants (Fatland E-mail address: [email protected] (A.J. Roger). et al., 2002), and filamentous fungi (Nowrousian et al., 2000), 1 These authors contributed equally to this work. 2 Current address: Department of Botany, University of British Columbia, Vancou- ACL enzyme activity requires ACLA, and ACLB (referred to here as ver, British Columbia, Canada. dual-subunit ACL, or dsACL) (Kanao et al., 2001). ACLA is http://dx.doi.org/10.1016/j.ympev.2015.05.010 1055-7903/Ó 2015 Elsevier Inc. All rights reserved. R.M.R. Gawryluk et al. / Molecular Phylogenetics and Evolution 91 (2015) 12–16 13 homologous to the b subunit of succinyl-CoA synthetase (SCS), (Yabuki et al., 2014), and numerous amoebozoans (e.g., Tubulinea while ACLB is homologous to the a subunit of SCS, fused to a small and a few Discosea; Fig. 1). It is therefore apparent that ssACL is portion homologous to citrate synthase (CS). An evolutionary not an animal-specific trait (Fatland et al., 2002). model for the origin of ACL from the aforementioned TCA cycle Hereafter, we reconstruct the evolutionary history of ACL, enzymes via gene duplication, fusion, and divergence has been updating and extending upon earlier analyses. We clarify the rela- suggested (Fatland et al., 2002). In contrast, animal ACL is a fusion tionships between dsACL and ssACL by carrying out in-depth phylo- protein (referred to here as single-subunit ACL, or ssACL) genetic analyses using a broad taxonomic sample of eukaryotes. (Elshourbagy et al., 1990, 1992); the N-terminal portion is homol- ogous to ACLA, and the C-terminal portion to ACLB. It was sug- 3.2. Eukaryotic ssACL and dsACL resolve into two monophyletic groups gested that ssACL represents a molecular synapomorphy of animals (Fatland et al., 2002), although ssACL homologs were We confirmed that all the subunits/domains constituting ACL recently identified in some non-ascomycete fungi (Hynes and apparently share the same evolutionary histories by performing Murray, 2010). separate phylogenetic analyses of homologs of: (1) ACLA and SCS Here, we have undertaken a comprehensive phylogenetic anal- b; (2) ACLB and SCS a; and (3) ACLB and CS (not shown). In agree- ysis of ACL across eukaryotes. We demonstrate that ssACL and ment with previous analyses (Fatland et al., 2002; Hügler et al., dsACL constitute ancient, distinct monophyletic lineages, that 2007), our Maximum Likelihood (ML) and Bayesian phylogenetic dsACL and ssACL have been laterally transferred and lost numerous reconstructions suggest that ACL homologs from green sulfur bac- times, and, contrary to previous analyses, that ssACL is likely the teria (i.e., Chlorobi) are basal to all others, and that bona fide ACL product of a gene fusion event that occurred very early in eukary- likely originated in this clade. Although the long-branch leading otic evolution. to this group does not preclude a long-branch attraction artifact, it is likely that ACL is ancestrally bipartite, and that ssACL, as observed in numerous eukaryotes, is a derived feature. 2. Methods To improve phylogenetic resolution, we concatenated ACLA and ACLB datasets, using Chlorobi as outgroup (Fig. 2). We found that 2.1. Taxon sampling and multiple alignment each of ssACL and eukaryotic dsACL constitute highly supported monophyletic groups, with the exception of ACL from ascomycetes ssACL, ACLA and ACLB homologs were identified via BLAST (discussed in Section 3.4). It is thus reasonable to conclude that queries of public databases (NCBI, Sanger Institute and MMETSP eukaryotes acquired ACLA and ACLB from a prokaryotic source – (Keeling et al., 2014)). In lineages with large numbers of ACL homo- although the backbone is not sufficiently resolved to infer the exact logs (e.g., fungi, and animals) a subset of phylogenetically repre- nature of the donor. Similarly, the low backbone resolution pre- sentative sequences was manually selected. Alignments were vents us from favoring an origin of ssACL through gene duplication generated with MAFFT L-INS-i v7 (Katoh and Standley, 2013). within the eukaryotic branch over a second LGT from prokaryotes Individual ACLA/ssACL and ACLB/ssACL alignments were concate- to eukaryotes before LECA. In either case, a single ancient gene nated, and trimmed automatically with BMGE 1.0, using the fusion event took place early in the eukaryotic line and led to the BLOSUM50 similarity matrix (Criscuolo and Gribaldo, 2010). establishment of ssACL. 2.2. Phylogenetics 3.3. ssACL and dsACL likely co-existed in LECA Maximum Likelihood (ML) phylogenies were estimated with Although ACL can be found in virtually all major eukaryotic RAxML version 8.0.19 (Stamatakis, 2014), under the groups, the patchy distribution of both ssACL and dsACL makes it PROTGAMMALGF model. Bootstrap support values estimated from difficult to ascertain precisely the timing of their emergence in 1000 replicates were mapped onto the estimated ML tree (obtained eukaryotes (Fig. 1). We suggest that both isoforms of ACL likely by 100 heuristic searches). arose early in eukaryotes, possibly prior to the last eukaryotic com- Bayesian analyses were carried out with PhyloBayes version mon ancestor (LECA). 3.3f (Lartillot et al., 2009) by running two chains under the catfix The internal relationships within main eukaryotic groups in the C20 + Poisson model, until convergence (maxdiff 0.1) after dis- ssACL clade are congruent with organismal phylogeny, and are carding 3000 burn-in trees.